ARTICLE

DOI: 10.1038/s41467-017-01427-1 OPEN Resistance to TGFβ suppression and improved anti- tumor responses in CD8+ T cells lacking PTPN22

Rebecca J. Brownlie 1,2, Celine Garcia1, Mate Ravasz1, Dietmar Zehn3, Robert J. Salmond 1,2 & Rose Zamoyska 1

Transforming growth factor β (TGFβ) is important in maintaining self-tolerance and inhibits reactivity. We show that CD8+ T cells that lack the tyrosine phosphatase Ptpn22, a major

1234567890 predisposing gene for autoimmune disease, are resistant to the suppressive effects of TGFβ. Resistance to TGFβ suppression, while disadvantageous in autoimmunity, helps Ptpn22−/− T cells to be intrinsically superior at clearing established tumors that secrete TGFβ. Mechanistically, loss of Ptpn22 increases the capacity of T cells to produce IL-2, which overcomes TGFβ-mediated suppression. These data suggest that a viable strategy to improve anti-tumor adoptive cell therapy may be to engineer tumor-restricted T cells with mutations identified as risk factors for autoimmunity.

1 Institute of Immunology and Infection Research, University of Edinburgh, Ashworth Laboratories, Charlotte Auerbach Road, Edinburgh EH9 3FL, UK. 2 Leeds Institute of Cancer and Pathology, University of Leeds, Wellcome Trust Brenner Building, St James’s University Hospital, Leeds LS9 7TF, UK. 3 School of Life Sciences Weihenstephan, Technical University Munich, Weihenstephaner Berg 3, Freising 85354, Germany. Correspondence and requests for materials should be addressed to R.J.S. (email: [email protected]) or to R.Z. (email: [email protected])

NATURE COMMUNICATIONS | 8: 1343 | DOI: 10.1038/s41467-017-01427-1 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01427-1

cell immune tolerance is maintained in the periphery by a response to both strong and weak , the latter being a Tcombination of extrinsic factors, including the action of common trait of tumor-associated antigens, which can otherwise regulatory T cells (Treg) and suppressive cytokines such as limit robust anti-tumor immune responses. These results suggest transforming growth factor β (TGFβ). TGFβ cytokines are posi- that targeting genes associated with susceptibility to auto- tive and negative regulators of proliferation and differentiation in immunity may be a viable strategy to improve the efficacy of many cell lineages, and play a key role in maintaining peripheral adoptive T cell immunotherapy. T cell tolerance1. TGFβ signaling was shown to be essential to 2–5 maintain peripheral T cell quiescence in vivo and to be a Results negative regulator of T cell proliferation in vitro6. Mice whose −/− Ptpn22 T cells resist TGFβ-mediated suppression.TGF-β T cells are unresponsive to TGFβ signals either by expressing a was shown previously to suppress low-affinity T cell responses more dominant negative TGFβRII transgene2,4, or through specificT effectively than high-affinity responses8 and we showed a similar cell deletion of the TGFβ receptor, Tgfbr13,5 die from fatal lym- role for PTPN22 in limiting CD8+ T cell responses18. Using the phoproliferative disease. In the absence of TGFβ signaling, per- − − OVA-specific TCR transgenic mouse, OT-1 on a Rag-1 / back- ipheral CD4+ and CD8+ T cells become activated in a cell- ground (hereafter called OT-1 T cells), for which a number of intrinsic fashion, most likely due to homeostatic expansion from peptides have been characterized, which span a range of affinities19, the lymphopenic environment shortly after birth, and develop − − we examined the sensitivity of control and Ptpn22 / OT-1 T cells effector function and pathogenicity7,8. Thus, TGFβ signals are to inhibition by TGFβ in vitro. particularly important for the regulation of the TCR-mediated, As reported previously8, TGFβ markedly inhibited - weak, self-ligand interactions that maintain peripheral T cell induced proliferation of OT-1 TCR transgenic T cells in a dose- homeostasis. dependent manner (Fig. 1a–c). This suppression occurred TGFβ signaling is important also in restricting T cell responses following stimulation both with strong agonist, SIINFEKL (N4) to tumors9,10. In the context of CD8+ T cells, TGFβ targets the peptide and with weak agonist, SIITFEKL (T4) peptide. Enhanced transcription of key effector molecules and interferes with their − − proliferation of Ptpn22 / OT-1 T cells compared to control OT- ability to kill tumor cells10. These observations have led to the 1 cells was particularly apparent in response to weak agonist suggestion that targeting TGFβ sensitivity, for example, by inac- peptide, T4, in the absence of TGFβ (Fig. 1a–c), as we reported tivating the TGFβ receptor in T cells, could be beneficial in anti- previously18. Significantly, both N4- and T4-induced proliferation tumor T cell therapy11. − − of Ptpn22 / OT-1 T cells were extremely refractory to TGFβ There are several parallels between the induction of T cell inhibition (Fig. 1a–c) compared to control OT-1 T cells. In autoimmunity and effective tumor immunosurveillance12,13. − − response to N4 peptide, no inhibition of Ptpn22 / OT-1 T cell T cells in autoimmune disease display functional characteristics proliferation was seen at doses of up to 5 ng/ml TGFβ, while in that, in a tumor setting, may be beneficial. For example, auto- − − response to T4 peptide, Ptpn22 / OT-1 T cells required ~10-fold reactive T cells respond effectively to poorly immunogenic anti- higher concentrations of TGFβ than control OT-1 T cells to show gens and are resistant to immune-regulatory mechanisms2,8. equivalent suppression of proliferation. Similarly, in order to mount effective anti-cancer responses, The mechanism of TGFβ-mediated suppression of T cell T cells need to overcome a strongly inhibitory tumor micro- responses is not well understood. PTPN22 is known to target environment, characterized by the presence of Tregs and high signaling molecules immediately downstream of the TCR, such as levels of inhibitory cytokines such as TGFβ. Indeed, TGFβ is − − Lck and ZAP70, so it was unclear why Ptpn22 / T cells would be expressed by a wide range of cancer cell types (https://portals. more resistant than control T cells to TGFβ-mediated suppres- broadinstitute.org/ccle/home) as well as by tumor-infiltrating sion. We asked at what point TGFβ suppression acted to inhibit immune cells such as myeloid-derived suppressor cells14. control OT-1 T cells. Time-course analysis demonstrated that Genome-wide association studies have shown that the hema- TGFβ inhibition of OT-1 T cell proliferation was apparent at day topoietic tyrosine phosphatase, PTPN22, has a commonly 3 (Fig. 1d), but in order to achieve this, TGFβ needed to be expressed variant, PTPN22R620W, that is highly associated with present from the start of culture. Delaying addition of TGFβ by predisposition to a number of autoimmune diseases15,16. PTPN22 2 days failed to limit antigen-induced proliferation (Fig. 1e). regulates Ag-specific T- and B-cell responses and some pattern Similarly, TGFβ-mediated inhibition of OT-1 T cell proliferation recognition signaling pathways in myeloid cells that lead to could be alleviated by an inhibitor to the TGFβ receptor, but only production of type 1 interferons15. We showed previously that − − when added from the beginning of culture (Fig. 1f). mouse Ptpn22 / T cells are partially resistant to wild-type Treg- These data show that TGFβ acted early in T cell activation, as mediated suppression17 and are more responsive to low-affinity by 48 h after stimulation cells were resistant to the anti- antigens18. Given the critical dual role of TGFβ in limiting proliferative effects of TGFβ. Accordingly, we analyzed activation autoimmune and anti-tumor responses, we investigated the marker and transcription factor expression at 24 h following impact of PTPN22 deficiency on the responses of CD8+ T cells to stimulation of OT-1 T cells. In control OT-1 T cells, TGFβ this key inhibitory cytokine. Furthermore, we sought to test the limited the extent of weak agonist T4-induced upregulation of the hypothesis that PTPN22 acts as a brake to limit the effectiveness − − IL-2 receptor-α chain (CD25), and key transcription factors Myc of anti-tumor T cell responses. We show that Ptpn22 / CD8 − − and IRF4, but less so Tbet, whereas Ptpn22 / OT-1 T cells were T cells are considerably more resistant to the suppressive effects refractory to these inhibitory effects of TGFβ (Fig. 2a, b). These of TGFβ than WT T cells. Concentrations of TGFβ that suppress data indicate that loss of PTPN22 counteracted TGFβ inhibition proliferation and differentiation of effector cytokines in WT − − during the early phase of TCR engagement. T cells show little inhibition of these processes in Ptpn22 / CD8 T cells. Upon stimulation with both weak and strong agonists − − −/− peptides Ptpn22 / CD8 T cells produce more IL-2 than WT Ptpn22 CD8+ T cells are superior at tumor clearance. The − − T cells and IL-2 interferes with the suppressive effect of TGFβ. resistance of Ptpn22 / T cells to TGFβ-mediated suppression − − Consequently upon adoptive transfer, Ptpn22 / CD8 T cells are in vitro led us to ask whether these CD8+ T cells would be more better able than WT CD8 T cells to control the growth of effective for adoptive cell therapy against TGFβ-secreting tumors − − established tumors that secrete TGFβ. Importantly, the superior in vivo. First, we evaluated the ability of control and Ptpn22 / − − anti-tumor capacity of Ptpn22 / CD8+ T cells is observed in OT-1 T cells to reject EL4 lymphoma cells (EL4-OVA) expressing

2 NATURE COMMUNICATIONS | 8: 1343 | DOI: 10.1038/s41467-017-01427-1 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01427-1 ARTICLE

abc0 β N4 N4 TGF (ng/ml) Control 2.00 Control 5 4 –/– –/– Ptpn22 1.00 Ptpn22 N4 T4 3 100 100 0.50 80 80 2 0.25 0.12 Control 60 60 1 0.06 40 40 0 0.03 5 5 0 20 20 0 0.2 0.6 1.7 0.2 0.6 1.7 0.06 0 0 0.06 2 3 4 5 2 3 4 5 010 10 10 10 010 10 10 10 T4 T4 100 100 4 1.00

/ml) 0.50 80 80 3 6 –/– 0.25 Ptpn22 60 60 2 0.12 40 40 0.06 20 20 1 Division index 0.03 Cell no. (x10 Events (% of max) 0 0 0 0.01 2 3 4 5 2 3 4 5 5 5 0 010 10 10 10 010 10 10 10 0 0.2 0.6 1.7 0.2 0.6 1.7 0.06 CTV 0.06 TGFβ (ng/ml)

d d1 d2 d3 100 100 100 80 80 80 0 TGFβ (ng/ml) 60 60 60 5 40 40 40 Events

(% of max) 20 20 20 0 0 0 0102 103 104 105 0102 103 104 105 0102 103 104 105 CTV

e TGFβ added: d0 d2 100 100 80 80 0 TGFβ (ng/ml) 60 60 5 40 40 Events

(% of max) 20 20 0 0 0102 103 104 105 0102 103 104 105 CTV

0 TGFβ (ng/ml) f TGFβ 5 DMSO: d0inhibitor: d0 d1 d2 100 100 100 100 80 80 80 80 60 60 60 60 40 40 40 40 Events

(% of max) 20 20 20 20 0 0 0 0 0102 103 104 105 0102 103 104 105 0102 103 104 105 0102 103 104 105 CTV

Fig. 1 Ptpn22 deficiency protects cells from TGFβ-mediated suppression. a Dilution of Cell Trace Violet (CTV) in control and Ptpn22−/− OT-1 cells was used to assess proliferation of cells stimulated with N4 or T4 peptide (10−6M) for 3 days in the presence of varying doses of TGFβ (0–5 ng/ml). b Proliferation was quantified by calculation of division index (FlowJo). c Numbers of live control and Ptpn22−/− OT-1 cells recovered after stimulation was calculated using MACSquant software (mean ± s.d. for triplicate replicates). d TGFβ inhibition becomes apparent only after 2d of culture. Dilution of CTV from control OT-1 cells stimulated with T4 (10−6 M) ± TGFβ (5 ng/ml) after d1, d2, or d3 of culture. e Dilution of CTV violet from control OT-1 cells stimulated with T4 (10−6 M) ± TGFβ (5 ng/ml) added at d0 or d2 of culture. f Inhibition of TGFβR1 signaling, by addition of ALK-5/TGFβR1 inhibitor SB431542, at d0 interferes with TGFβ-mediated suppression of proliferation, but SB431542 is less effective when added on d1 or d2 of culture. Control OT-1 cells stimulated with T4 (10−6 M) ± TGFβ (5 ng/ml) were cultured for 3d ± 5 μM SB431542 (or DMSO vehicle control) added on d0, d1, or d2 of culture. Representative histograms of CTV dilution analysed by flow cytometry are shown in a, d, e, f. Data are representative of 5 (a–c), 2 (d, e), and 3 (f) separate experiments high-affinity N4 peptide as a tumor-specific antigen. EL4-OVA TGFβ is secreted by EL4 cells, so we asked to what extent TGFβ – tumors were established subcutaneously for 5d before transfer of was preventing tumor rejection by OT-1 T cells2,20 22. Sub- − − small numbers (5 × 104/mouse) of either control or Ptpn22 / cutaneous tumors were established as before, but with an EL4- naive OT-1 T cells. Tumor size was monitored for a further 7d OVA cell line engineered to express a soluble form of the after which mice were killed to prevent control animals exceeding TGFβRII (EL4-OVA STβRII)10. Adoptively transferred control permitted tumor volumes, and wet tumor mass measured. OT-1 T cells were better able to reduce tumor growth when TGFβ − − − − Strikingly, Ptpn22 / OT-1 T cells were significantly superior to was blocked, as shown previously10, while Ptpn22 / OT-1 T cells control cells in controlling growth of subcutaneous EL4-OVA effected complete rejection of EL4-OVA STβRII cells (20/20 tumors (*p < 0.05 using two-way ANOVA with Tukey’s post test) mice) (Fig. 3b). These data illustrate that while, not completely − − − − (Fig. 3a). In particular, very few animals receiving Ptpn22 / OT- immune to TGFβ inhibition, Ptpn22 / T cells are inherently 1 T cells showed evidence of tumor mass over the time course superior in managing growth of TGFβ-secreting tumors. following transfer, whereas a number of those receiving control Key to control of tumor mass is the ability of CD8+ T cells to OT-1 T cells showed ongoing tumor growth followed by regres- kill tumor targets. TGFβ acts to inhibit the expression of several sion toward the end of the experiment. cytolytic gene products, thus hindering CD8 T cell cytotoxicity10.

NATURE COMMUNICATIONS | 8: 1343 | DOI: 10.1038/s41467-017-01427-1 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01427-1

a No stim T4 – TGFβ T4 + TGFβ 100 100 100 100 80 80 80 80 Control 60 60 60 60 40 40 40 40 20 20 20 20 0 0 0 0 0102 103 104 105 0102 103 104 105 0102 103 104 105 0102 103 104 105 CD25 cMyc IRF4 Tbet 100 100 100 100 80 80 80 80 –/– Ptpn22 60 60 60 60 40 40 40 40 20 20 20 20 Events (% of max) 0 0 0 0 0102 103 104 105 0102 103 104 105 0102 103 104 105 0102 103 104 105 CD25 cMyc IRF4 Tbet

b 200 * * *** *** Control **** **** ** 150 Ptpn22–/–

100

MFI (normalized) 50

0 TGFβ: – +–+–+–+ CD25 cMyc IRF4 Tbet

Fig. 2 PTPN22 mediates the extent of TGFβ-mediated suppression of early activation markers. a Control and Ptpn22−/− OT-1 cells were cultured for 24 h with T4 peptide (10−6 M) ± TGFβ (5 ng/ml). Activation was assessed by upregulation of CD25 and transcription factors c-Myc, IRF4, Tbet by flow cytometry. Representative histograms are shown. b Quantification of MFIs of peptide stimulated cells ± TGFβ 5 ng/ml as shown in a (mean and s.d. of three replicates and data representative of 5 experiments). *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001, as determined using two-way ANOVA with Tukey’s post test for multiple comparisons

To better mimic responses to weak tumor antigens, we used ID8 luciferase (Fluc), enabling us to monitor bioluminescence as a murine ovarian carcinoma cells23 that as a tumor-specific antigen correlate of tumor mass. Tumors were established in the expressed the weak agonist peptide, T4, which is ~70-fold less peritoneum of B6 mice prior to adoptive transfer of control or − − potent in stimulating OT-1 T cells than N4 peptide24, and is of Ptpn22 / OT-1 T cells, that were first activated in vitro. Tumor equivalent affinity to self-peptides that select the thymic bioluminescence was analyzed 14d post T cell transfer. While repertoire19. We cultured ID8-T4 cells with control and Ptpn22 control OT-1 T cells reduced tumor mass in a number of − − / OT-1 T cells in vitro and monitored production of granzyme recipient mice, this effect did not reach statistical significance − − B (GzB) and IFNγ. Remarkably, while GzB was detectable in less (Fig. 3e). By contrast, adoptive cell transfer of Ptpn22 / OT-1 − − than 10% of control cells, ~70% of Ptpn22 / T cells were positive cells resulted in a significant decrease in luminescence signal, and for GzB and a substantial proportion of the latter also expressed in the majority of mice, a complete rejection of tumor load IFNγ in response to ID8-T4 cells alone. Blockade of TGFβ (Fig. 3e, *p < 0.05 using two-way ANOVA with Tukey’s post test). signaling by addition of a TGFβ-receptor inhibitor relieved the Taken together, our data using EL4 lymphoma and ID8 − − suppression of control T cells so that now ~40% expressed ovarian carcinoma models indicate that Ptpn22 / T cells − − granzyme B while >90% of Ptpn22 / cells became GzB+ generate a more effective anti-tumor responses to both high- (Fig. 3c). and low-affinity TSA. This effect most likely stems from a − − We then compared the ability of control and Ptpn22 / OT-1 combination of the enhanced TCR responsiveness and prolifera- CD8+ TCR transgenic T cells to respond to tumors expressing tion of cells lacking PTPN22 to both high- and low-affinity TSA low-affinity TSA in vivo. ID8-T4 tumors were established in the and their reduced susceptibility to inhibitory cytokines such as peritoneum of recipient B6 (CD45.1/CD45.2F1) mice for 28d TGFβ, permitting differentiation and expression of effector before transfer of an equal mixture of control CD45.1+ and cytokines and the cytolytic machinery. − − Ptpn22 / CD45.2+ OT-1 T cells. Three days later, we found − − significantly enhanced proliferation of Ptpn22 / OT-1 T cells compared to control cells in individual mice (Fig. 3d, **p < 0.01 Canonical TGFβR signaling is unaffected by loss of PTPN22. using two-way ANOVA with Tukey’s post test) showing that We next assessed the mechanisms underlying the ability of − − − − Ptpn22 / T cells indeed proliferate more in response to the weak Ptpn22 / T cells to overcome TGFβ-mediated suppression. T4 antigen expressed by the tumor. Surface TGFβRII expression (Fig. 4a) and canonical TGFβRII In order to non-invasively assess ID8 tumor growth in vivo, we signaling, judged by TGFβ-induced phospho-SMAD (Fig. 4b), − − transduced ID8-T4 cells with lentivirus-expressing Firefly were identical in control and Ptpn22 / OT-1 T cells.

4 NATURE COMMUNICATIONS | 8: 1343 | DOI: 10.1038/s41467-017-01427-1 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01427-1 ARTICLE

adControl

EL4-OVA ) –/– 3 1500 Ptpn22 2.5 ** *** * 100 * 100 2.0 0.3 80 80 1000 1.5 60 0.2 60 1.0 40 500 40

0.1 Division index 20 20 0.5 Events (% of max) Tumor weight (g)

% of mice with tumors 0 0.0 Tumor mean volume (mm 0 0 0.0 2 3 4 5 Cells: – –/– – –/– 5791112 010 10 10 10 –/– Day CTV 2 Control Control Control Ptpn22 Ptpn22 Ptpn2

β beEL4-OVAsT Rll ) 3 1500 Pre treatment Post treatment

**** /sr) 2 *** 100 Cont – 0.3 **** 1000

80 Radiance *** (p/s/cm 0.2 60 x108 40 500 0.1 3 20 Ptpn22 Tumor weight (g) % of mice with tumors 0.0 0 Tumor mean volume (mm 0 2

Cells: – –/– – –/– 5791114

Day –/– Control Control 1 Ptpn22 Ptpn22

–/– Control – Control Ptpn22 c ID8 + Ptpn22 –/– **** ns * –/– 11 11 11 ID8 + OT1 Ptpn22 OT1 100 10 10 10 **** 10 10 10 5 8.41 0.85 5 56.76 11.09 10 10 10 10 10 80 No TGF **** 109 109 109 4 4 inhibitor 10 10 60 8 8 8 10 10 10 3 3 7 7 7 10 10 40

% cells 10 10 10 β

R 6 6 6 2 2 20 10 10 10 10 10 granzyme B +ve Toltal flux (p/s) 5 5 5 0 0 10 10 10 90.14 0.60 30.52 1.63 0 4 4 4 2 3 4 5 2 3 4 5 TGFβR inhibitor 0 10 10 10 10 0 10 10 10 10 – + 10 10 10 Pre Post Pre Post Pre Post 38.65 1.06 82.65 11.39 105 105

+ TGFbR 5000 *** inhibitor 104 104 4000 3 3 10 10 3000 102 102 2000 0 0 60.22 0.06 5.89 0.07 2 3 4 5 2 3 4 5 1000

0 10 10 10 10 0 10 10 10 10 Grazyme B MFI Granzyme B (of positive cells) 0 γ IFN TGFβR inhibitor – +

Fig. 3 Loss of Ptpn22 improves the response of CD8 T cells to tumors. a, b B6 mice were injected s.c. with 1 × 106 EL4-OVA (a) or EL4-OVAsTβRII cells (b). Five days later, 5 × 104 control or Ptpn22−/− OT-1 cells were injected i.v. Tumor weight (left panel) and tumor incidence (middle panel) is shown (d12 for EL4-OVA, d14 for EL4-OVAsTβRII from two pooled experiments, n = 20/group). Right panel shows mean tumor volume of individual mice, which received no exogenous T cells (black lines), control OT-1 T cells (blue lines), or Ptpn22−/− OT-1 T cells (red lines) until termination of the experiment on d12 (a)or d14 (b) from one experiment (n = 10/group). c Control and Ptpn22−/− OT-1 cells were co-cultured with ID8 tumor cells ± ALK-5/TGFR1 inhibitor SB431542. After 2 days, Brefeldin A was added to cultures for 6 h and cells were stained for intracellular granzyme B and IFNγ analysis by flow cytometry. Representative dot plots and quantification (mean and s.d. of three replicates) are shown. a–c *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 as determined using two-way ANOVA with Tukey’s post test for multiple comparisons and for incidence data ***p < 0.001 and ****p < 0.0001 as determined by χ2-test with Fisher’s exact post test adjusting the p value for number of comparisons. Data are representative of three separate experiments. d Groups (n = 6) of CD45.1/2 B6 mice were injected i.p. with 1 × 107 ID8 cells, followed 28 days later with an i.p. injection of a mix of CTV-labeled control CD45.1+ OT-1 and Ptpn22−/− CD45.2+ OT-1 cells (1 × 106 of each). Peritoneal lavage was harvested 3 days later and assessed for OT-1 T cell proliferation by flow cytometry. Division index was calculated using Flowjo software. **P < 0.01 as determined by Student’s paired t test. Data are representative of two experiments. e Groups (n = 7) of B6 mice were injected i.p. with 5 × 106 ID8-T4-fluc2 cells and assessed for tumor establishment on d11 (pre treatment) by bioluminescence imaging. On d12, groups of mice received either no cells, 10 × 106 control OT-1 CTLs, or Ptpn22−/− OT-1 CTLs i.p. All mice were assessed for tumor growth by bioluminescence imaging on d26 (14 d post treatment). Example images of n = 4 mice/group are shown d11 and d26 (pre/post treatment). Graphs show bioluminescence signal intensity of all mice on d11 and d26 (pre/post treatment). *p < 0.05, ****p < 0.0001, as determined by Student’s paired t test. Data are representative of two experiments

Furthermore, in both genotypes, there was SMAD-dependent PTPN22 substrates ZAP70 and TCRζ26 following TCR stimula- inhibition of granzyme B expression20 and upregulation of CD44 tion of control OT-1 T cells in the presence or absence of TGFβ. expression by high-dose TGFβ (Fig. 4c). Recent analyses have Importantly, TGFβ did not impair basal or TCR-stimulated levels shown that expression of the transcription factor FoxP1 is of Zap70 Y493, TCRζ Y83, or ERK MAPKinase phosphoryation essential for TGFβ-mediated suppression of anti-tumor T cells25, (Fig. 4e). Previous data have shown that PTPN22 associates with − − however, levels of FoxP1 expression by control and Ptpn22 / Csk, which may be important for its function in T cells27. T cells were comparable under basal and activated conditions However, co-immunoprecipitation of PTPN22 with Csk was (Fig. 4d). unaffected by culture of OT-1 T cells with TGFβ (Fig. 4f). In order to determine whether TGFβ directly affects PTPN22 Together, these data show that PTPN22 does not impact directly function, we analyzed the phosphorylation of well-defined upon canonical SMAD-dependent TGFβ receptor signaling

NATURE COMMUNICATIONS | 8: 1343 | DOI: 10.1038/s41467-017-01427-1 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01427-1

abdControl Control Control Isotype No TGFβ ns Ptpn22–/– Ptpn22–/– Ptpn22–/– Control Control Control TGFβRll 5 ng/ml TGFβ N4 Ptpn22–/– Ptpn22–/– Ptpn22–/– 15,000 Control 100 –/– 100 100 Ptpn22 ns 80 80 80 ns 10,000 60 60 60

40 40 40 FoxP1 MFl 5000 Events (% of max) Events (% of max) 20 Events (% of max) 20 20

0 0 0 0 0102 103 104 105 0102 103 104 105 0102 103 104 105 N4 –++ TGFβ ––+ TGFβRll pSMAD2/3 FoxP1

c Control e N4 No TGFβ –/– 100 30 Ptpn22 150 N4 + TGFβ Control Control β 5 ng/ml TGF Ptpn22–/– Ptpn22–/– 20 100 100 100 50 80 75 10 60 pZap70 Y493 (MFI) 50 pTCR zeta Y83 (MFl) 40 50 0 pErk(T202, Y204)+ cells (%) 0 0 2 5 10 0 2 5 10 0 2 5 10 Events (% of max) 20 25 Time (mins) 0 0 0102 103 104 105 0 0.06 1.7 0.6 0.2 5 f ip Ptpn22 GranzymeB Ptpn22 100 kD 100 250

80 200

60 150 Csk 50 kD 40 100 TGFβ – +

Events (% of max) 20 50 % of control sample 0 0 0102 103 104 105 0 0.06 1.7 0.6 0.2 5 CD44 TGFβ (ng/ml)

Fig. 4 TGFβ signaling is not affected by the absence of PTPN22. Histogram overlays show equivalent expression in control and Ptpn22−/− OT-1 cells of a cell surface TGFβRII and b intracellular SMAD2/3 phosphorylation (pSMAD) after stimulation with 5 ng/ml TGFβ for 30 min. c The TGFβ-regulated genes granzyme B and CD44 were down- and upregulated, respectively, in both control and Ptpn22−/− OT-1 cells by TGFβ in a dose-dependent manner after activation with N4 10−8 M TGFβ for 24 h. d FoxP1 induction was comparable in the presence of N4 10−8 M in control and Ptpn22−/− OT-1 cells ±5 ng/ml TGFβ for 24 h. e Intracellular phosphorylation of Zap70, TCR zeta and ERK in OT-1 cells after TCR stimulation with N4 10−6 M ± 5ng/ml TGFβ for 0–10 min. f Co-immunoprecipitation of Ptpn22 and Csk in OT-1 cells pretreated ± TGFβ for 2.5 h. Data are from one experiment representative of at least three experiments (mean and s.d. of three replicates). Non significance (ns) as determined using two-way ANOVA with Tukey’s post test for multiple comparisons pathways or FoxP1-dependent processes, as these are unaffected common gamma chain cytokines on TGFβ suppression has − − in Ptpn22 / T cells, and conversely that TGFβ does not directly been reported7,30. We found that, compared with control cells, − − affect the function of PTPN22 in regulating T cell signaling. Ptpn22 / OT-1 T cells secreted significantly greater quantities of IL-2 in response to both strong N4 and weak T4 agonist peptides (Fig. 5c, ***p < 0.001 and ****p < 0.0001 using two-way ANOVA IL-2 protects against TGFβ-mediated suppression with Tukey’s post test). Importantly, TGFβ treatment almost . Our data − − suggested that indirect mechanisms were likely to underlie the completely blocked control but not Ptpn22 / OT-1 T cell IL-2 − − ability of Ptpn22 / T cells to resist TGFβ inhibition. Serendipi- production (Fig. 5c). −/− Two distinct approaches confirmed the importance of elevated tously, we noted that when control and Ptpn22 OT-1 T cells − − were activated with T4 peptide as a 1:1 mix in the same culture TCR-induced IL-2 production for the ability of Ptpn22 / T cells wells, the inhibitory effect of TGFβ on control T cell proliferation to resist TGFβ-mediated inhibition. First, canonical IL-2- − − was abrogated (Fig. 5a). These data suggested that Ptpn22 / dependent signaling pathways were blocked using the inhibitor T cells secreted or expressed a factor that “rescued” control cells tofacitinib, which inhibits Jak3 in addition to other members of from the effects of TGFβ in a co-culture system. Culture of the Jak kinase family. Tofacitinib markedly reduced the ability of − − − − control and Ptpn22 / T cells in transwells also abrogated Ptpn22 / OT-1 T cells to withstand TGFβ-mediated inhibition, β indicating that IL-2 cytokine receptor signaling was important for the inhibitory effect of TGF on control T cell proliferation, − − indicating that a soluble factor was responsible for the the resistance of Ptpn22 / cells to TGFβ (Fig. 5d). Second, the protective effect (Fig. 5a, b). TGFβ was shown to limit T cell IL-2 addition of recombinant IL-2 to T cell cultures prevented the production20,28,29 and a protective effect of IL-15 and other inhibitory effects of TGFβ on OT-1 T cell proliferation when

6 NATURE COMMUNICATIONS | 8: 1343 | DOI: 10.1038/s41467-017-01427-1 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01427-1 ARTICLE added at either d0 or on d2 (Fig. 5e). Recovery of proliferation that elevated IL-2 production in response to TCR triggering − − was dose dependent (Fig. 5f) confirming that insufficiency of IL-2 enables Ptpn22 / T cells to overcome the inhibitory effects of production in the presence of TGFβ accounted for the inhibition TGFβ. of OT-1 T cell proliferation. Together, these data strongly suggest

abc– TGFβ Control 1.5 + TGFβ Ptpn22–/– –/– * Control Ptpn22 N4 T4 100 100 1.6 **** 0.8 80 80

/ml) 1.0 60 60 6 **** **** 1.2 0.6 40 40 *** 20 20 *** Sep 0.8 0.4 **** **** 0 0 0.5 2 3 4 5 2 3 4 5

010 10 10 10 010 10 10 10 Cell no. (x10 **** IL-2 (ng/ml) 0.4 0.2 100 100 80 80 0.0 0.0 0.0 60 60 –6 –8 –6 –8 –6 –8 –6 –8 Mix –+–+–+TGFβ 10 10 10 10 10 10 10 10 40 40 5 ng/ml TGFβ 5 ng/ml TGFβ 20 20 Sep Mix TW 0 0 0102 103 104 105 0102 103 104 105 100 100 80 80 TW 60 60 40 40 e 20 20 – TGFβ + TGFβ Events (% of max) 0 0 +lL-2: None d0 d2 0102 103 104 105 0102 103 104 105 100 100 100 CTV 80 80 80 60 60 60 d –/– Control Ptpn22 Control 40 40 40 2.00 **** *** *** 20 20 20 1.00 0 0 0 *** 0102 103 104 105 0102 103 104 105 0102 103 104 105 /ml) 6 0.50 100 100 100 0.25 80 80 80

–/– Events (% of max) 60 60 60 0.12 Ptpn22 40 40 40

Cell no. (x10 0.06 20 20 20 0.03 0 0 0 2 3 4 5 2 3 4 5 2 3 4 5 ––++ ––++ TGFβ 010 10 10 10 010 10 10 10 010 10 10 10 – + – + Jak3i CTV

f – TGFβ + TGFβ g IL-2: 0 4 16 62.5 250 1000 pg/ml –/– 100 Control Ptpn22 150 80 ns 60 **** Control 40 100 20 0 0102 103 104 105 0102 103 104 105 0102 103 104 105 0102 103 104 105 0102 103 104 105 0102 103 104 105 50

100 80

Events (% of max) 0 Nuclear NEATc1 (normalized) β 60 β –/– T4 Ptpn22 40 T4 20 No stim No stim T4 TGF 0 T4 TGF 0102 103 104 105 0102 103 104 105 0102 103 104 105 0102 103 104 105 0102 103 104 105 0102 103 104 105 CTV

Fig. 5 Increased IL-2 production from Ptpn22−/− OT-1 cells overcomes TGFβ-mediated suppression. a Proliferation of CTV-labeled control and Ptpn22−/− OT-1 cells stimulated with 10−8 M N4 for 3 days ±TGFβ (5 ng/ml) cultured in separate wells, in the same well (mix) or in transwells. Representative histograms showing CTV dilution are shown. b Quantification of recovered control cells. Bars represent mean ± s.e.m. of six individual animals pooled from two separate experiments, *p < 0.05 as determined using Students t test. c IL-2 was quantified by ELISA in supernatants of control and Ptpn22−/− OT-1 cells stimulated for 24 h with N4 or T4 (10−6,10−8 M) in the presence of anti-CD25 Ab (10 μg/ml) TGFβ (5 ng/ml). Bars represent triplicate values from groups of three animals, error bar represents s.d. d Jak3 signaling is required to overcome TGFβ-mediated suppression. Control and Ptpn22−/− OT-1 cells stimulated with N4 (10−6 M) ± TGFβ (5 ng/ml) were cultured for 3d with or without JAK3 inhibitor added for the last 24 h of culture. Live cell recovery is shown, dots represent triplicate values of pooled mice, line represents mean and error bars s.d. e Supplementation with IL-2 as late as d2 of culture can override TGFβ-mediated suppression. Control and Ptpn22−/− OT-1 cells stimulated with T4 peptide (10−6 M) for 3d, ±TGFβ (5 ng/ml) and ±IL-2 (1 ng/ml) added on d0 or d2. Representative histograms of CTV dilution are shown. f Reversal of TGFβ-mediated suppression by titrating IL-2. Control and Ptpn22−/− OT-1 cells stimulated with N4 (10−6 M) ±TGFβ (5 ng/ml) were cultured for 3d either without or with the addition of exogenous IL-2 at varying concentrations (0–1000 pg/ml). Representative histograms of CTV dilution are shown. Data c–f is representative of at least three experiments. ***p < 0.001 and ****p < 0.0001, as determined using two-way ANOVA with Tukey’s post test for multiple comparisons. g Ptpn22−/− OT-1 cells are resistance to TGFβ inhibition of NFATc1 translocation. Control and Ptpn22−/− OT-1 cells stimulated with T4 (10−6 M) ±TGFβ for 4 h were lysed and nuclear extracts analysed for NFATc1 expression. Data are pooled and normalized from three separate experiments. Error bars represent s.e.m. and ****p < 0.0001, as determined using Student’s t test

NATURE COMMUNICATIONS | 8: 1343 | DOI: 10.1038/s41467-017-01427-1 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01427-1

Previous studies have suggested that TGFβ interferes with inflammatory response. Thus, the effects of these two cytokines TCR-induced Ca2+ signaling and subsequent NFAT activation, in vivo are likely to be highly context dependent. thereby limiting T cell activation and IL-2 production31. We have The mechanisms by which TGFβ inhibits TCR signaling and T shown previously that Ca2+ fluxes in naive T cells were not cell function are not fully understood. An early report suggested influenced by the absence of Ptpn2217. However, in vitro that, in CD4+ T cells, TGFβ signals interfere with very early TCR- measurement of Ca2+ flux in primary naive T cells by FACS induced Tek and ERK kinase activation and NFAT nuclear requires very acute and strong stimulation, e.g., in the OT-1 translocation31. Our data support this view as we saw clear model by using multivalent peptide:MHC complexes containing inhibition of NFAT translocation in control cells incubated with − − strong agonist peptide N432. Under these stimulation conditions, TGFβ, whereas Ptpn22 / T cells were resistant to this inhibiiton. we saw no reduction of Ca2+ signals in control OT-1 T cells in the Analysis of more upstream pathways did not reveal more speci- presence of TGFβ (Supplementary Fig. 1). However, we reasoned fically how TGFβ inhibition intersects TCR signaling, as we saw that looking more downstream, e.g., at NFAT nuclear transloca- no change in TCRζ, Zap70, or MAPK signaling in WT T cells tion might reveal some differences. Indeed, we observed stimulated in the presence or absence of TGFβ. However, these significant inhibition of NFAT nuclear translocation in the assays utilize strong stimulation of the TCR over very short time presence of TGFβ when control OT-1 T cells were stimulated frames and may therefore lack the sensitivity to observe TGFβ with T4 peptide (Fig. 5g, ****p < 0.0001 using Student’s t test). In inhibition. More recently, TGFβ-mediated T cell suppression has contrast, T4-induced NFATc1 nuclear translocation in stimulated been linked to effects on the mTOR pathway37. In those studies, − − Ptpn22 / CD8 T cells was remarkably resistant to suppression suppressive effects of TGFβ on mTOR signaling were assessed by TGFβ (Fig. 5g). These data corroborate the finding that TGFβ following 36 h (or more) of T cell stimulation, time points at suppresses control CD8 T cell proliferation by interfering with IL- which autocrine IL-2 signals contribute to mTOR activation38. 2 production through inhibition of NFAT activation, and indicate Therefore, it is possible that TGFβ inhibition of mTOR signals at that in the absence of PTPN22, this inhibitory pathway is these later time points might, at least in part, be explained by circumvented. reduced IL-2 secretion. In NK cells, TGFβ also suppresses mTOR activation with potent downstream effects on cellular metabo- lism39. Given the importance of shifts in cellular metabolism in Discussion the generation and differentiation of effector T cells, it will be The ability of T cells to integrate and respond appropriately to a interesting to determine whether TGFβ also impedes these pro- complex network of signals triggered by antigen receptor cesses in T cells. A recent study indicated that FoxP1 expression engagement, cytokines, chemokines, and other environmental was required for TGFβ-mediated inhibition of anti-tumor CD8+ − − signals is critical to the outcome of immune responses. Further- T cells25. However, our data showed that Ptpn22 / and control more, in order to maintain tolerance, cell-intrinsic inhibitory OT-1 T cells expressed equivalent levels of FoxP1. Therefore, it is mechanisms and the action of suppressive cytokines, including likely that FoxP1 expression is necessary but not sufficient to TGFβ prevent T cell activation by low-affinity self-antigens. In mediate TGFβ suppression. the current work, we found that altering the sensitivity of TCR Our experiments provide new insight into the mechanisms by signaling, by deletion of the autoimmune-associated phosphatase which PTPN22 dampens T cell activation. Loss of PTPN22 PTPN22, further impacted upon the sensitivity of CD8+ T cells to function increases the chance of T cells responding to weak, self- the suppressive effects of TGFβ. Our results describing the con- antigens and this enhanced TCR sensitivity is compounded by a sequent imbalance in positive and negative signaling have knock-on effect on the inhibitory TGFβ pathway. While these important implications for our understanding of the mechanisms effects are likely to be detrimental in the context of autoimmune regulating autoimmune T cells and may also have practical disease, they also raise the possibility that GWAS-identified, applications in tumor therapy. In this regard, we showed that the autoimmune susceptibility loci can be manipulated to improve combination of enhanced TCR sensitivity and subsequent resis- the intrinsic efficacy of adoptive cell therapy (ACT) for cancer. In − − − − tance to TGFβ inhibition enabled Ptpn22 / T cells to have this regard, we showed that Ptpn22 / T cells were superior to potent anti-tumor effector function in vivo. control cells in responding to both high- and low-affinity TSA A key mechanism underlying resistance to TGFβ suppression and in mediating tumor rejection in vivo, suggesting that tar- − − was the ability of Ptpn22 / T cells to produce IL-2 more geting PTPN22 may be a viable approach in T cell immu- abundantly than control T cells. Other studies have shown that notherapies. We reason that in future studies, by restricting loss − − Ptpn22 / CD4+ T cells make more cytokines, including IL-2, of PTPN22 to tumor-specific T cells, undesirable autoimmunity − − than WT T cells33,34 and we show that is true for Ptpn22 / will be minimized. Despite the recent success of chimeric CD8+ cells also. Previous data indicated that Smad3-dependent antigen receptors40, a major advantage of retaining TCR inhibition of Il2 transcription and protein expression was key to expression for ACT is that many TSAs are intracellular proteins TGFβ-induced suppression of TCR-induced T cell proliferation35. that are only presented in the context of MHC. However, TSAs Consistent with this, in our studies, suppression of peptide- are frequently too weak to stimulate effective T cell responses, but induced control OT-1 T cell proliferation could be reversed by by manipulating the intracellular signaling machinery, we addition of recombinant IL-2 to cultures. It is worthy of note that show that it is possible to harness traits that are undesirable in PTPN22 does not directly regulate IL-233 or TGFβ receptor sig- autoreactive T cells, such as reactivity to weak antigens and naling pathways. Therefore, it is likely that the ability of Ptpn22 resistance to suppressive cytokines, for the benefit of anti-tumor − − / T cells to withstand TGFβ-induced suppression is a result of responses. quantitative effects of elevated IL-2 production rather than qua- β litative effects on IL-2R or TGF R signaling. In addition, our data Methods − − − − − − highlight the complexity of the interplay between IL-2 and TGFβ Mice. Rag-1 / OT-I CD45.1+/+ 40, Ptpn22 / Rag-1 / OT-I CD45.2+/+ 18 and − − signaling on T cell responses. In this regard, IL-2 abrogates the CD45.1+/ CD45.2+/ mouse strains were bred and maintained under specific and pro-inflammatory effects of TGFβ on CD4+ Th17 differentiation opportunistic pathogen-free conditions at the Universities of Edinburgh and Leeds. 36 Age-matched (6–12 weeks) and sex-matched mice were used in all experiments. and instead favors inducible Treg expansion . By contrast, our fi + Animal procedures were approved under a UK Home Of ce project license and data show that elevated IL-2 production by CD8 T cells abro- were performed in compliance with the ethical guidelines of the Universities of gates suppressive effects of TGFβ and instead favors a pro- Edinburgh and Leeds.

8 NATURE COMMUNICATIONS | 8: 1343 | DOI: 10.1038/s41467-017-01427-1 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01427-1 ARTICLE

Flow cytometry and antibodies. The following conjugated antibodies were used: ratio of 398 nm (Indo-1 bound to Ca2+)/482 nm (unbound Ca2+) in OT-1 cells by phycoerythrin–indotricarbocyanine–anti-CD8β (clone eBioH35-17.2, cat flow cytometry and all data were acquired on an LSRII (BD Bioscience). 15266777, 1:400 dilution), phycoerythrin–anti-CD25 (clone PC61.5, cat 15298489, 1:200 dilution), allophycocyanin–eFluor 780–anti-CD44 (clone IM7, cat 15351620, fl – αβ Cell lines. ID8 ovarian carcinoma cells were a gift from Dr K. Roby, University of 1:400 dilution), uorescein isothiocyanate anti-TCR (clone H57-597, cat 11- 23 – Kansas, USA . EL4-OVA cells were provided by Dr K. Okkenhaug, Babraham 5961-81, 1:200 dilution), phycoerythrin anti-T-bet (clone 4B10, cat 12-5825-80, β 1:100 dilution), eFluor 450–anti-IRF4 (clone 3E4, cat 15519846, 1:200 dilution), Institute, UK and EL4-OVAsTGF RII cells were provided by Prof. J. Massagué, eFluor 450–anti–granzyme B (clone NG2B, cat 15371830, 1:100 dilution), (all from Memorial Sloan Kettering Cancer Institute, New York, USA. All cell lines were eBioscience); Alexa Fluor 488–anti-CD25 (clone PC61.5, cat 102018, 1:200 dilu- negative for mouse pathogens, including mycoplasma contamination (IMPACT tion), allophycocyanin–anti-CD45.1 (clone A20, cat 110713, dilution 1:200), IV, IDEXX BioResearch). fluorescein isothiocyanate–anti-CD45.1 (clone A20, cat 110718, dilution 1:200), phycoerythrin–anti-CD45.2 (clone 104, cat 109807, dilution 1:200), Brilliant Violet ID8 tumor experiments. ID8 ovarian carcinoma cells23 were transduced with 421–anti-CD45.2 (clone 104, cat 109832, dilution 1:200), Alexa Fluor 488–anti- retroviral constructs to express the ova-variant protein T4 (aminoacids 197–387). IFN-γ (clone XMG1.2, cat 505813, dilution 1:200) (all from BioLegend). Uncon- ID8-T4 cells were maintained in IMDM supplemented with 10% FCS, 50 μM2- jugated rabbit antibodies to c-Myc (clone D84C12, cat 5605, dilution 1:50), mercaptoethanol, L-glutamine and antibiotics (100 U penicillin, 100 mg/ml strep- phospho-Smad2 (Ser465/467)/Smad3 (Ser423/425) (clone D27F4, cat D27F4, tomycin, 0.5% ciproxin). For co-culture experiments, sub-confluent ID8-T4 cells dilution 1:100), FoxP1 (clone D35D10 XP, cat 4402, dilution 1:200), phospho- were harvested and transferred to 48-well tissue culture plates at 2 × 104 cells/well. Zap70 (Tyr493)/Syk (Tyr526) Y493 (cat 2704, dilution 1:100), phospho-p44/42 Cells were allowed to settle and adhere to culture plates for 4 h then OT-I T cells MAPK (Erk1/2) (Thr202/Tyr204) (clone D13.14.4E, cat 4370, dilution 1:100) were were added (2 × 105/well) and cells co-cultured for 48 h. For analysis of T cell from Cell Signaling Technologies and anti-CD3 zeta (phospho Y83) (clone EP776 intracellular cytokine expression, Brefeldin A (2.5 mg/ml, Sigma) was added to (2)Y, cat ab68236, dilution 1:50) was from Abcam; all were counterstained with wells for the last 6 h of culture prior to staining using the antibodies described goat anti-rabbit (A21244; Molecular Probes). Phycoerythrin anti-mouse TGFbeta above. RII (cat FAB532P, 1:50) and isotype control were from R&D systems. Live/Dead For in vivo T cell expansion studies, abdominal tumors were established using − − Aqua and Cell Tracer Violet (CTV) dyes were from Life Technologies. For intra- 107 ID8 cells injected i.p. to recipient CD45.1+/ CD45.2+/ mice. After 28 days, − − cellular staining, cells were fixed in Phosflow Lyse/Fix Buffer 1 (BD) or FoxP3 Fix/ OT-I CD45.1+/+ and Ptpn22 / OT-I CD45.2+/+ T cells were mixed 1:1, stained Permeabilization Buffer (eBioscience) before staining with antibody in Permeabi- with CTV dye (Molecular Probes) and transferred i.p. to tumor-bearing or control lization/Wash buffers (eBioscience). For phospho-SMAD2/3, staining cells were mice. After 3 days, peritoneal lavages were performed and tumor-induced T cell fixed with 2% paraformaldehyde and permeabilized with methanol prior to proliferation assessed as described above. staining. Samples were acquired with a MACSQuant flow cytometer (Miltenyi) and For tumor rejection studies, ID8-T4 cells were stably transduced with Firefly data were analyzed with FlowJo software (Treestar). Division index is the average luciferase-expressing lentiviral vector (kindly provided by M. Lorger, University of number of cell divisions that a cell in the original population has undergone (i.e., Leeds) (ID8-T4-fluc2). Abdominal tumors were established by injection of 5 × 106 includes the undivided peak). ID8-T4-fluc2 cells i.p. into B6 recipients. Quantification of tumor growth was performed via non-invasive bioluminescence imaging using IVIS Spectrum and Living Image software (Perkin Elmer) on day 11 prior to adoptive T cell T cell culture and stimulation. T cells from the lymph nodes of wild-type and transfer. T cells for transfer were generated in vitro by stimulating OT-1 control or −/− ’ fi ’ − − Ptpn22 mice were cultured in Iscove s Modi ed Dulbecco s Medium (IMDM, Ptpn22 / cells for 2 days with 0.01 μM N4, washed, then their populations μ Invitrogen) supplemented with 10% FCS, L-glutamine, antibiotics and 50 M2- expanded for a further 4 days in 20 ng/ml recombinant human IL-2 (Peprotech). − − mercaptoethanol. The agonist peptide SIINFEKL (N4) and partial agonist SIIT- On day 12, 107 OT-1 control or OT-1 Ptpn22 / effector CTLs cells were injected i. FEKL (T4) (Peptide Synthetics) were added to culture medium at concentrations p. into recipient tumor-bearing mice and tumor growth monitored by fi β stated in gure legends. Recombinant human TGF (R&D) was added to cultures bioluminescence imaging at various time points thereafter. at 5 ng/ml unless otherwise stated and, where indicated, a TGFβ inhibitor (SB431542, 2.5 μM, Abcam), Jak3 inhibitor (tofacitinib, 200 nM, Tocris Bioscience), or recombinant human IL-2 (Peprotech) were also added. Transwell EL4 tumor experiments. EL4-OVA and EL4-OVAsTGFβRII cells were main- experiments were performed using HTS Transwell 96 system 0.4 μm pore poly- tained in IMDM supplemented with 10% FCS, 50 μM 2-mercaptoethanol, L- carbonate membrane (Corning). For measurement of IL-2, cells were stimulated glutamine and antibiotics (100 U/ml penicillin, 100 μg/ml streptomycin, 400 μg/ml with peptide ± TGFβ in the presence of blocking CD25 mAb (Biolegend) in order geneticin (G418) and 0.5% ciproxin). Tumor cells in log-phase growth were to prevent IL-2 consumption. IL-2 in culture supernatants was measured using the resuspended in PBS and 106 cells were injected subcutaneously into the right dorsal Mouse IL-2 ELISA Ready-SET-Go!® kit (eBioscience). flank. Expression of SIINFEKL/H-2Kb pMHC complexes in >97% EL4 cells was verified by flow cytometry using anti-SIINFEKL/H-2Kb antibody (25 D1.6). Tumor growth was monitored at least twice weekly using calipers with tumor volume Immunoprecipitation and western blot. T cells were generated in vitro by sti- calculated as length × width × the average of length and width measurement to − − mulating OT-1 control or Ptpn22 / cells for 2d with 0.01 μM N4, washed, then approximate the total volume of the subcutaneous tumor mass (mm3)41. Five days expanded for a further 6d in 20 ng/ml recombinant human IL-2 (Peprotech) fol- after tumor cell injection, a time point at which 30–50% of animals presented with lowed by treatment for 2.5 h ± TGFβ (5 ng/ml). Cells were lysed in 1% TritonX- palpable tumor mass, recipient mice were randomly divided into three groups. 100, 0.5% n-dodecyl-b-D-maltoside, 50 mM Tris-HCl, 150 mM NaCl, 20 mM Control mice received no adoptive cell transfer, whereas other groups received − − EDTA, 1 mM NaF, and 1 mM sodium orthovanadate containing protease inhibi- adoptive transfer of 105 control or Ptpn22 / OT-1 T cells by intravenous injec- tors. Samples were immunoprecipitated with anti-Ptpn22 antibody (clone D6D1H, tion. Following 7–10 days, mice were killed, tumors resected and tumor mass Rabbit mAb Cell Signaling Technology) coupled to Dynabeads prot G (10003D; determined. Life Technology). Western blots were performed on samples and membranes were probed with anti-Ptpn22 rabbit mAb (clone D6D1H, cat CS14693S, dilution 1:1000 Statistical analysis ’ from Cell Signaling Technology) and anti-murine CSK antibody (clone 52, cat . Prism software was used for Student s t test (paired or unpaired; two tails), two-way ANOVA with Tukey’s post test for multiple com- 610079, dilution 1:1000, from BD Transduction Labs) and proteins detected with χ2 ’ secondary Abs and visualized using infrared imaging system (Odyssey; LI-COR parisons and -test with Fisher s exact post test. Biosciences). For uncropped western blot image, see Supplementary Fig. 2. Data availability. The data that support the findings of this study are available − − within the article and its Supplementary Information files and from the corre- NFAT translocation assay. Control and Ptpn22 / OT-1 cells (2 × 106 cells) were − sponding authors upon reasonable request. stimulated with T4 peptide (T4 6 M) ± TGFβ for 4 h at 37 °C. Cells were then lysed and nuclear extracts prepared and analysed for NFATc1 using the TransAM NFATc1 Transcription Factor assay kit according to the manufacturer’s instruc- Received: 14 December 2016 Accepted: 14 September 2017 tions (Activemotif).

Calcium flux. OT-1 cells (1 × 107 cells) were prestained with anti-CD8β (clone eBioH35-17.2, cat 15266777, 1:400 dilution) in order to identify cell populations prior to loading with 2 µM Indo-1 AM (Life Technologies) for 40 min at 37 °C in PBS. Cells were resuspended in pre-warmed IMDM containing 1% FCS and References 2+ β supplemented with 2.5 mM CaCl2 and 2.8 mM MgCl2. The baseline Ca levels 1. Rubtsov, Y. P. & Rudensky, A. Y. TGF signalling in control of T cell-mediated were measured for 1 min before PE-labeled N4 dextramer (Immudex, Denmark) self-reactivity. Nat. Rev. Immunol. 7, 443–453 (2007). was added to stimulate the cells. Some samples were pretreated with TGFβ (5 ng/ 2. Gorelik, L. & Flavell, R. A. Abrogation of TGFβ signaling in T cells leads to ml) 5 min before the addition of dextramer. Ionomycin (1 µg/ml) was added as a spontaneous T cell differentiation and autoimmune disease. Immunity 12, positive control for measuring saturated Ca2+ levels. Data are represented as the 171–181 (2000).

NATURE COMMUNICATIONS | 8: 1343 | DOI: 10.1038/s41467-017-01427-1 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01427-1

3. Li, M. O., Wan, Y. Y. & Flavell, R. A. T cell-produced transforming growth 31. Chen, C. H. et al. Transforming growth factor blocks tec kinase factor-beta1 controls T cell tolerance and regulates Th1- and Th17-cell phosphorylation, Ca2+ Influx, and NFATc translocation causing inhibition of T differentiation. Immunity 26, 579–591 (2007). cell differentiation. J. Exp. Med. 197, 1689–1699 (2003). 4. Lucas, P. J., Kim, S. J., Melby, S. J. & Gress, R. E. Disruption of T cell 32. Daniels, M. A. & Jameson, S. C. Critical role for CD8 in T cell receptor binding homeostasis in mice expressing a T cell-specific dominant negative and activation by peptide/major histocompatibility complex multimers. J. Exp. transforming growth factor beta II receptor. J. Exp. Med. 191, 1187–1196 Med. 191, 335–346 (2000). (2000). 33. Hasegawa, K. et al. PEST domain-enriched tyrosine phosphatase (PEP) 5. Marie, J. C., Liggitt, D. & Rudensky, A. Y. Cellular mechanisms of fatal early- regulation of effector/memory T cells. Science 303, 685–689 (2004). onset autoimmunity in mice with the T cell-specific targeting of transforming 34. Maine, C. J., Teijaro, J. R., Marquardt, K. & Sherman, L. A. PTPN22 contributes growth factor-beta receptor. Immunity 25, 441–454 (2006). to exhaustion of T lymphocytes during chronic viral infection. Proc. Natl Acad. 6. McKarns, S. C. & Schwartz, R. H. Distinct effects of TGF-beta 1 on CD4+ and Sci. USA 113, E7231–E7239 (2016). CD8+ T cell survival, division, and IL-2 production: a role for T cell intrinsic 35. McKarns, S. C., Schwartz, R. H. & Kaminski, N. E. Smad3 is essential for TGF- Smad3. J. Immunol. 174, 2071–2083 (2005). beta 1 to suppress IL-2 production and TCR-induced proliferation, but not IL- 7. Johnson, L. D. S. & Jameson, S. C. TGF-β sensitivity restrains CD8+ T cell 2-induced proliferation. J. Immunol. 172, 4275–4284 (2004). homeostatic proliferation by enforcing sensitivity to IL-7 and IL-15. PLoS ONE 36. Laurence, A. et al. Interleukin-2 signaling via STAT5 constrains T helper 17 cell 7, e42268 (2012). generation. Immunity 26, 371–381 (2007). 8. Zhang, N. & Bevan, M. J. TGF-β signaling to T cells inhibits autoimmunity 37. Delisle, J.-S. et al. The TGF-β-Smad3 pathway inhibits CD28-dependent cell during lymphopenia-driven proliferation. Nat. Immunol. 13, 667–673 (2012). growth and proliferation of CD4 T cells. Genes Immun. 14, 115–126 (2013). 9. Gorelik, L. & Flavell, R. A. Immune-mediated eradication of tumors through 38. Cornish, G. H., Sinclair, L. V. & Cantrell, D. A. Differential regulation of T cell the blockade of transforming growth factor-beta signaling in T cells. Nat. Med. growth by IL-2 and IL-15. Blood 108, 600–608 (2006). 7, 1118–1122 (2001). 39. Viel, S. et al. TGF-beta inhibits the activation and functions of NK cells by 10. Thomas, D. A. & Massagué, J. TGF-β directly targets functions repressing the mTOR pathway. Sci. Signal. 9, ra19 (2016). during tumor evasion of immune surveillance. Cancer Cell 8, 369–380 (2005). 40. Maude, S. L. et al. Chimeric antigen receptor T cells for sustained remissions in 11. Won, J. et al. Tumorigenicity of mouse thymoma is suppressed by soluble type leukemia. N. Engl. J. Med. 371, 1507–1517 (2014). II transforming growth factor beta receptor therapy. Cancer Res. 59, 1273–1277 41. Helmich, B. K. & Dutton, R. W. The role of adoptively transferred CD8 T cells (1999). and host cells in the control of the growth of the EG7 thymoma: factors that 12. Joseph, C. G. et al. Association of the autoimmune disease scleroderma with an determine the relative effectiveness and homing properties of Tc1 and Tc2 immunologic response to cancer. Science 343, 152–157 (2014). effectors. J. Immunol. 166, 6500–6508 (2001). 13. Maueroder, C. et al. Tumor immunotherapy: lessons from autoimmunity. Front. Immunol. 5, 212 (2014). Acknowledgements 14. Gabrilovich, D. I. & Nagaraj, S. Myeloid-derived suppressor cells as regulators – We thank J. Massagué for providing EL4-OVA STβRII cells and K. Okkenhaug for of the immune system. Nat. Rev. Immunol. 9, 162 174 (2009). fl 15. Bottini, N. & Peterson, E. J. Tyrosine phosphatase PTPN22: multifunctional providing EL4-OVA cells; M. Lorger and T. Andreou for providing Fire y luciferase- regulator of immune signaling, development, and disease. Annu. Rev. Immunol. expressing lentiviral vector, Judi Allen and Paul Travers for critically reading the 32,83–119 (2014). manuscript; David Wright and Xiaoyan Zou for technical assistance; Graeme Cowan and 16. Rhee, I. & Veillette, A. Protein tyrosine phosphatases in lymphocyte activation Margo Chase-Topping for statistical help. The work was funded by Wellcome Trust and autoimmunity. Nat. Immunol. 13, 439–447 (2012). Investigator Award, WT096669AIA, to R.Z., a strategic award, WT095831, for the Centre 17. Brownlie, R. J. et al. Lack of the phosphatase PTPN22 increases adhesion of for Immunity, Infection and Evolution, and a Cancer Research UK grant (23269) to R.S. murine regulatory T cells to improve their immunosuppressive function. Sci. Signal. 5, ra87 (2012). Author contributions 18. Salmond, R. J., Brownlie, R. J., Morrison, V. L. & Zamoyska, R. The tyrosine R.J.B., R.J.S., and R.Z. designed the study; R.J.B., C.G., M.R., and R.J.S. carried out the phosphatase PTPN22 discriminates weak self peptides from strong agonist TCR experiments; D.Z. developed and supplied key reagents; R.J.B., R.J.S., and R.Z. wrote the signals. Nat. Immunol. 15, 875–883 (2014). manuscript; R.J.S. and R.Z. contributed key concepts and share equal senior authorship. 19. Daniels, M. A. et al. Thymic selection threshold defined by compartmentalization of Ras/MAPK signalling. Nature 444, 724–729 (2006). 20. Gorelik, L. & Flavell, R. A. Transforming growth factor-beta in T cell biology. Additional information Nat. Rev. Immunol. 2,46–53 (2002). Supplementary Information accompanies this paper at doi:10.1038/s41467-017-01427-1. 21. Massague, J. TGFbeta in cancer. Cell 134, 215–230 (2008). fi fi 22. Zhang, Q. et al. Blockade of transforming growth factor-{beta} signaling in Competing nancial interests: The authors declare no competing nancial interests. tumor-reactive CD8(+) T cells activates the antitumor immune response cycle. Mol. Cancer Ther. 5, 1733–1743 (2006). Reprints and permission information is available online at http://npg.nature.com/ 23. Roby, K. F. et al. Development of a syngeneic mouse model for events related to reprintsandpermissions/ ovarian cancer. Carcinogenesis 21, 585–591 (2000). 24. Zehn, D., Lee, S. Y. & Bevan, M. J. Complete but curtailed T cell response to Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in fi very low-affinity antigen. Nature 458, 211–214 (2009). published maps and institutional af liations. 25. Stephen, T. L. et al. Transforming growth factor beta-mediated suppression of antitumor T cells requires FoxP1 transcription factor expression. Immunity 41, 427–439 (2014). 26. Wu, J. et al. Identification of substrates of human protein-tyrosine phosphatase Open Access This article is licensed under a Creative Commons PTPN22. J. Biol. Chem. 281, 11002–11010 (2006). Attribution 4.0 International License, which permits use, sharing, 27. Cloutier, J. F. & Veillette, A. Association of inhibitory tyrosine protein kinase adaptation, distribution and reproduction in any medium or format, as long as you give p50csk with protein tyrosine phosphatase PEP in T cells and other hemopoietic appropriate credit to the original author(s) and the source, provide a link to the Creative cells. EMBO J. 15, 4909–4918 (1996). Commons license, and indicate if changes were made. The images or other third party 28. Brabletz, T. et al. Transforming growth factor beta and cyclosporin A inhibit material in this article are included in the article’s Creative Commons license, unless the inducible activity of the interleukin-2 gene in T cells through a indicated otherwise in a credit line to the material. If material is not included in the noncanonical octamer-binding site. Mol. Cell Biol. 13, 1155–1162 (1993). article’s Creative Commons license and your intended use is not permitted by statutory 29. Das, L. & Levine, A. D. TGF-beta inhibits IL-2 production and promotes cell regulation or exceeds the permitted use, you will need to obtain permission directly from cycle arrest in TCR-activated effector/memory T cells in the presence of the copyright holder. To view a copy of this license, visit http://creativecommons.org/ sustained TCR signal transduction. J. Immunol. 180, 1490–1498 (2008). licenses/by/4.0/. 30. Sanjabi, S., Mosaheb, M. M. & Flavell, R. A. Opposing effects of TGF-beta and IL-15 cytokines control the number of short-lived effector CD8+ T cells. Immunity 31, 131–144 (2009). © The Author(s) 2017

10 NATURE COMMUNICATIONS | 8: 1343 | DOI: 10.1038/s41467-017-01427-1 | www.nature.com/naturecommunications